Figure 14.35(a) shows the residual plot versus x for the simple linear regression of the laptop service time data. This plot fans out, indicating that the error term e tends to become larger in magnitude as x increases. To remedy this violation of the constant variance assumption, we divide all terms in the simple linear regression model by x. This gives the transformed model

Data from Figure 14.35 (a)

Transcribed Image Text:

2 = B₁ (²7) + X + B₁ + E X or, equivalently,= Bo + Bil -B + A(-²) + B₁ X